Posts
Comments
Ape trounces the best of the human world in memory competition
In 1999 we hadn't experienced the dotcom crash or 911. Those events may have slowed consumer technology application by a few years. On the other hand, military robots, remotely piloted aircraft, and "social" software for tracking terrorist groups have seen accelerated development. Concerns about global warming and oil shortages will likely accelerate nanotech and biotech associated with energy production while reducing the pace of development in other fields. Computational power continues to increase by a factor of a thousand per decade. Biotech continues to advance exponentially.
If humanity were really approaching a technological singularity I'd expect to see rapid increases in average wealth. Stock market performance in the last decade doesn't reflect a growth in real wealth. Also death rates for common diseases aren't showing a significant decline.
Memories decay exponentially. This occurs both over time and over number of items to remember. Also, remembering requires brain attention. The vast majority of memories in a brain will never be activated sufficiently for conscious awareness. As memories accumulate, the fraction we actively access decreases.
The human mind is a flashlight that dimly illuminates the path behind. Moving forward, we lose sight of where we've been. Living a thousand years wouldn't make that flashlight any brighter.
Terren Suydam: "So genetics is not the whole story, and that's what I mean by group selection."
I use the term "multilevel selection" for what you are describing. I agree it has been important.
E.g., there has been selection between different species. Species with genomes that supported rapid adaptation to changing environments and that supported quick diversification when expanding into new niches spread far and wide. (Beetles have been extremely successful with around 350,000 known species.) Other specie branches died out. The genetic mechanisms and the animal body plans that persist to the present are the winners of a long between specie selection process.
My intuition is that selection operating at the individual level, whether genetic or cultural, suffices to produce cooperation and moral behavior. Multilevel selection probably played a supporting role.
Terren Suydam: "The first is that one has to adopt the group-selection stance."
(Technical jargon nitpick.)
In studying evolutionary biology, "group-selection" has a specific meaning, an individual sacrifices its own fitness in order to improve the group fitness. I.e., individual loss for a group gain. E.g., suppose you have a species that consists of many small family groups. Suppose a mutation produces a self-sacrificing individual in one of the groups. His fitness is slightly lower but his family group fitness is higher. His group tends to grow faster than other groups. So his group produces more splinter groups, some of which will have his alleles. Within any one group his allele tends to die out, but the overall population frequency of the allele increases due to the increased number of splinter groups containing the allele. This is an example of group-selection.
Much more common is cooperation that doesn't lower the individual's fitness. In this case it is win-win, individual gain and group gain. Symbiosis is an example where the cooperation is between different species. Both individuals gain so it is not an example of group-selection.
There are a few known examples of group-selection but they tend to be the rare exception, not the rule. Often something appears to be group-selection but on closer analysis turns out to be regular selection. E.g., suppose a hunter shares his meat with the tribe. He isn't lowering his fitness because he already has enough meat for himself. He is publicly displaying his prowess as a food provider which increases his mating success. Thus his generosity directly improves his fitness. His generosity is a fitness increasing status display.
Cooperation can and usually does arise through regular selfish selection.
(I see EY also noted this.)
Caledonian: "It is much, much more elegant - and more compatible with what we know about cognition - to hold that the complex systems are built out of smaller, simpler systems over which the complex has no control."
The brain has feedback loops to even the earliest processing stages. Thus, I might choose to look for a lost contact lens. With that goal in mind, my unconscious visual processing systems will be primed to recognize signals that could be a contact lens. (The feedback loops can be observed in the neural tissue. There are cognitive science experiments that demonstrate that high level conscious decisions can affect neural processing in the earlier stages.)
The conscious mind may be a dim reflection of the top level computation that makes choices but it does reflect some of the processing that occurs. The conscious mind is aware of possible future outcomes and potential paths to preferred outcomes. The conscious mind isn't aware of the total brain mechanism that makes decisions, but it is aware of important pieces of that computation.
EY: "human cognitive psychology has not had time to change evolutionarily over that period"
Under selective pressures, human populations can and have significantly changed in less than two thousand years. Various behavioral traits are highly heritable. Genghis Khan spread his behavioral genotype throughout Asia. (For this discussion this is a nitpick but I dislike seeing false memes spread.)
re: FAI and morality
From my perspective morality is a collection of rules that make cooperative behavior beneficial. There are some rules that should apply to any entities that compete for resources or can cooperate for mutual benefit. There are some rules that improved fitness in our animal predecessors and have become embedded in the brain structure of the typical human. There are some rules that are culture specific and change rapidly as the environment changes. (When your own children are likely to die of starvation, your society is much less concerned about children starving in distant lands. Much of modern Western morality is an outcome of the present wealth and security of Western nations.)
As a start I suggest that a FAI should first discover those three types of rules, including how the rules vary among different animals and different cultures. (This would be an ongoing analysis that would evolve as the FAI capabilities increased.) For cultural rules, the FAI would look for a subset of rules that permit different cultures to interact and prosper. Rules such as kill all strangers would be discarded. Rules such as forgive all trespasses would be discarded as they don't permit defense against aggressive memes. A modified form of tit-for-tat might emerge. Some punishment, some forgiveness, recognition that bad events happen with no one to blame, some allowance for misunderstandings, some allowance for penance or regret, some tolerance for diversity. Another good rule might be to provide everyone with a potential path to a better existence, i.e., use carrots as well as sticks. Look for a consistent set of cultural rules that furthers happiness, diversity, sustainability, growth, and increased prosperity. Look for rules that are robust, i.e., give acceptable results under a variety of societal environments.
A similar analysis of animal morality would produce another set of rules. As would an analysis of rules for transactions between any entities. The FAI would then use a weighted sum of the three types of moral rules. The weights would change as society changed, i.e., when most of society consists of humans then human culture rules would be given the greatest weight. The FAI would plan for future changes in society by choosing rules that permit a smooth transition from a human centered society to an enhanced human plus AI society and then finally to an AI with human origins future.
Humans might only understand the rules that applied to humans. The FAI would enforce a different subset of rules for non-human biological entities and another subset for AI's. Other rules would guide interactions between different types of entities. (My mental model is of a body made up of cells, each expressing proteins in a manner appropriate for the specific tissue while contributing to and benefitting from the complete animal system. Rules for each specific cell type and rules for cells interacting.)
The transition shouldn't feel too bad to the citizens at any stage and the FAI wouldn't be locked into an outdated morality. We might not recognize or like our children but at least we wouldn't feel our throats being cut.
"...if there was a pill that would make the function of the mirror neurons go away, in other words, a pill that would make you able to hurt people without feeling remorse or anguish, would you take it?"
The mirror neurons also help you learn from watching other humans. They help you intuit the feelings of others which makes social prediction possible. They help communication. They also allow you to share in the joy and pleasure of others...e.g., a young child playing in a park.
I would like more control over how my mind functions. At times it would be good to turn-off some emotional responses, especially when someone is manipulating my emotions. So if the pill had only temporary effects, it were safe, and it would help me achieve my goals then, yes, I'd take the pill.
roko: "Game theory doesn't tell you what you should do, it only tells you how to do it. E.g. in the classic prisoner's dilemma, defection is only an optimal strategy if you've already decided that the right thing to do is to minimize your prison sentence."
Survival and growth affect the trajectory of a particle in mind space. Some "ethical systems" may act as attractors. Particles interact, clumps interact, higher level behaviors emerge. A super AI might be able to navigate the density substructures of mind space guided by game theory. The "right" decision would be the one that maximizes persistence/growth. (I'm not saying that this would be good for humanity. I'm only suggesting that a theory of non-human ethics is possible.)
(Phil Goetz, I wrote the above before reading your comment: "...variation in possible minds, for sufficiently intelligent AIs, is smaller than the variation in human minds" Yes, this what I was trying to convey by "attractors" and navigation of density substructures in mind space.)
"Are there morally justified terminal (not instrumental) values, that don't causally root in the evolutionary history of value instincts?"
Such a morality should confer survival benefit. E.g., a tit-for-tat strategy.
Suppose an entity is greedy. It tries to garner all resources. In one-on-one competitions against weaker enties it thrives. But other entities see it as a major threat. A stronger entity will eliminate it. A group of weaker entities will cooperate to eliminate it.
A super intelligent AI might deduce or discover that other powerful entities exist in the universe and that they will adjust their behavior based on the AI's history. The AI might see some value in displaying non-greedy behavior to competing entities. I.e., it might let humanity have a tiny piece of the universe if it increases the chance that the AI will also be allowed its own piece of the universe.
Optimal survival strategy might be a basis for moral behavior that is not rooted in evolutionary biology. Valued behaviors might be cooperation, trade, self restraint, limited reprisal, consistency, honesty, or clear signaling of intention.
HA, Ben Jones
I appreciate the compliment and your interest in my views, however, for now, I would rather read what others have to say on this topic.
Vassar: "The neuropsychology of illusory decision procedures however is disturbing to a different disposition than the existence of a future."
Yes. HA's point about neuroscience and the illusion of "I" is largely orthogonal to EY's discussion concerning choice and determinism. However, the neuroscience that HA references is common knowledge in EY's peer group and is relevant to the topic under discussion...so why doesn't EY respond to HA's point?
(Consider an experiment involving the "hollow face" illusion. The mind's eye sees an illusionary face surface. However when asked to touch the mask nose subjects move their finger directly to the sunken surface, the subjects don't hesitate at the illusionary surface. The brain has multiple internal visual representations. Our internal "I" has no direct assess to the visual representation used to direct the finger motion. (One visual pathway goes from the occipital lobes at the rear of the brain upward through the dorsal regions to the frontal lobes. Another visual pathway goes from the occipital lobes downward through the ventral regions and guides the finger movement. The "mind's eye" only has direct access to the information passing dorsally.)
Conscious awareness is only a dim reflection of the brain's computational operation. "I" is a poor model of the human mind.
For this discussion I use "consciousness" to refer to the mind's internal awareness of qualia. Consciousness may be an inherent property of whatever makes up the universe, i.e., even individual photons may have some essence of consciousness. Human type consciousness might then arise whenever sufficient elements group together in the right pattern. Other groupings into other patterns might generate other types of consciousness. Consciousness may have no purpose. Or perhaps certain types of consciousness somehow enhance intelligence and provide an evolutionary advantage.
If I don't trust that other people have a self awareness much like mine, then I have no reason to trust any of my senses or memories or beliefs. So I trust the evidence that other humans look like me, act like me, have brains like mine, and express internal thoughts in language as I do. I am only slightly less certain that mammals, birds, reptiles, and fish are conscious as they share common ancestry, have similar brain structures, and exhibit similar behavior. I am less certain about insects or worms. As I don't know the physical correlates of consciousness, the further from myself an entity is in structure and behavior, the less certain I am that it has an internal awareness similar to my own.
Animal consciousness can be explored by experimentation on humans, primates, mice, and fruit flies. The boundaries of consciousness can be mapped in the neural tissue of the brain. Cognitive scientists can explore what stimuli provoke a conscious response, what provoke an unconscious response, and what don't provoke any response. Scientists can observe what brain tissue is active when we say we experience qualia and what is active when we say we don't experience qualia. Studying brain injury patients provides a wealth of information concerning the brain's generation of consciousness...split brain, phantom limbs, aphasias, personality changes, delusions, etc.
Such experimentation indicates that our internal concept of self is largely an illusion. The mind tries to make sense out of whatever is available. If both brain hemispheres are strongly connected then there is a strong illusion of one internal person. If the brain hemispheres are disconnected, then experiments show two different personalities inhabiting the same brain. Each personality has no awareness of the other personality. When the second personality acts independently, the first personality rationalizes why the first personality "chose" to perform the action. It is possible that many such self aware personalities co-exist in our brains, each with its own illusion of being in control and each with its own perception of qualia. (In some brain injuries, a person no longer believes that their own arm is part of self. Even though they can control the arm and feel what the arm touches, they think it is someone else's arm. The brain function that creates the illusion of self is broken.) These illusions of self may not be necessary to experience qualia but probably are necessary for a human to describe or relate the experience of qualia.
Speculation about zombies should take into account what science has already discovered. I.e., our internal concept of ourselves is only a blurred reflection of reality. "Self" is manufactured on the fly out of bits and pieces that change with every experience, with every hormonal change, with every drug we take, or with every injury we experience.
What would our internal "self" experience as each neuron were gradually replaced by a nano computer simulator? If the simulator generated a similar essence of qualia (i.e., simulating a brain pattern is sufficient to generate the experience of qualia) then the internal experience should be the same. If the simulator produced no such experience of qualia, then our internal self would be unable to recognize that our internal awareness was shrinking. We would not be able to remember that we could once hear more sounds or see more colors as memory itself depends on that internal awareness. Our internal self would fade away unnoticed by that internal self. (In some cases of dementia, the patient doesn't comprehend that his mind is failing. He don't understand why his family has brought him to see the doctor.) With nano-simulators mental function would continue, but internal awareness might disappear.
"So now we have a group of scientists who set out to test correlation A, but found correlation B in the data instead. Should they publish a paper about correlation B?"
Since you testing multiple hypotheses simultaneously, it is not comparable to Eliezer's example. Still, it is an interesting question...
Sure. The more papers you publish the better. If you are lucky the correlation may hold in other test populations and you've staked your claim on the discovery. Success is largely based on who gets credit.
Should a magazine publish papers reporting correlations with relatively high P-values? When thousands of scientists are data mining for genetic correlations to disease, chance correlations will be very common. If the genetic difference occurred in a metabolic pathway known to be relevant to the disease, the correlation might be publishable even with a high P-value. If the scientists just reported a random correlation they should have a low P-value.
A better approach might be to replace publication in a journal by some other mechanism. Suppose there were an online, centralized database for hypotheses related to a disease or trait. No single population study would be meaningful, but multiple reports by different researchers in different populations would be significant. Evidence would accumulate and credit would be shared among all those responsible for validating or disproving the hypothesis.
Caledonian: "We have no end of fools who feel one way or another. How about giving people credit for doing what they think is right? Or even better, what they can demonstrate to be correct?"
I don't believe a person with an IQ around 125 and the skill to get elected POTUS is a fool. I respect intelligence and knowledge but those are not the only or even the most important traits necessary for leadership.
I don't really want to defend Bush. I just don't find him any worse than Clinton, Kerry, or Gore. I was also curious to see the reaction to my post. I found the concept of "Happy Death Spirals" interesting and wondered if it would be demonstrated on this thread.
"One application: If you find yourself in a group of people who tell consistently unfunny jokes about the Hated Enemy, it may be a good idea to head for the hills, before you start to laugh as well..."
Another step on the path to hermit mountain.
Robin: "Would jokes where Dilbert's pointy-headed boss says idiotic things be less funny if the boss were replaced by a co-worker? If so, does that suggest bosses are Hated Enemies, and Dilbert jokes bring false laughter?"
Not really...consider, "The inmates are running the asylum.", i.e., clueless idiots are in charge and ruining our lives. When a co-worker is an idiot and is ruining his own life it is just pathetic.
Aaron: "Even the part about Bush being a big dumb brute isn't in huge contention..."
I don't mind when a brilliant scientist calls Bush dumb, but I find it ironic when, as is often the case, the person calling Bush dumb has an IQ below 120, is scientifically illiterate, and has no achievement comparable to POTUS. Clearly a high IQ scientist will respect neither Bush's intelligence nor his scientific knowledge...I feel pretty much the same about politicians of all flavors. Given his innate limits, I give Bush some credit for doing what he feels is right, too many politicians seem motivated only by personal benefit.
Stable population of asexual haploid bacteria considering only lethal mutations:
Let "G" be the genome string length in base pairs.
Let "M" be the mutations per base pair per division.
Let "numberOfDivisions" be the average number divisions a bacterium undergoes before dying.
Let "survivalFraction" be the probability that division produces another viable bacterium.
survivalFraction = (1 - M)**G. (Assuming mutation events are independent.)
1 = numberOfDivisions x survivalFraction. (Assuming population size is stable.)
Then ln(1/numberOfDivisions) = G ln(1 - M).
G = -ln(numberOfDivisions) / ln(1 - M).
Using Taylor series for small M gives
ln(1 - M) = -M + higher order terms of M.
So G = ln(numberOfDivisions) / M.
Which does not match the simulation observation that G = O(1 / M**2).
Summarizing my thoughts:
1) For lethal mutations the rule, "one mutation, one death", holds.
In life few mutations will be lethal. Even fewer in a sexual species with genetic redundancy. So the information content limits calculated by assuming only lethal mutations will not apply to the human genome.
2) Selection may not directly affect population size.
E.g., in sexual selection winners and losers are balanced so the total number of offspring is relatively constant. So minor harmful mutations may be removed with high efficiency without affecting total population size.
3)High selection pressure may drive the specie gene pool high up a local fitness peak. However being "too optimized" might hurt specie survival by lowering variance and making the specie more vulnerable to environmental variation, e.g., new pathogens. Or it may decrease the probability of a two-mutation adaptation that might have improved competitiveness again a different species. (Humans may eventually out-compete fruit flies.)
4) Working with selection, crossover and assortative mating remove the most harmful mutations quickly at a high "death" cost (worst case is one death per mutation removal) and remove less harmful mutations slowly at a low "death" cost. The "mutation harmfulness" vs. "mutation frequency" graph likely follows a power law. It should be possible to derive a "mutation removal efficiency" relationship for each "mutation fitness cost". Such functions are likely different for each specie and population structure.
5) Selection operates on traits. Traits usually depend on complex network interaction of genetic elements. Most genetic elements simultaneously affect many traits. Therefore most trait values will follow an inverted bathtub curve, i.e., low and high values are bad and the mid-range is good. (Body homeostasis requires stable temperature, ph, oxygen level, nutrient level, etc.) Evolution has favored robust systems with regulatory feedback to adjust for optimal trait values in the face of genetic, stochastic, and environmental variation.
(The "bath tub curve" is essentially a one-state system. Multi-state regulatory systems are also common in biology and can be used to differentiate cells.)
6) Total genome information content is limited by the mutation rate and the number of bit errors that are removed by selection. (In the Shannon sense of a message being a string of symbols from a finite set and transmission between generations being a noisy communication channel.) I believe this numerical limit is highly dependent on specie reproductive biology and population dynamics.
Increases in genome information content are not directly related to "evolutionary progress". In evolution the genome "meaning" is more important than the genome "message". Over evolutionary time the average "meaning" value of each bit may be increasing. Evolution of complex genetic regulatory systems increased the average "meaning" value per bit. Evolution of complex brains capable of "culture" increased the average "meaning" value per bit. (The information bits that give humans the ability to read are more valuable than the information bits in a book.)
7) The total information in a specie genome can be far greater than the information contained in any individual genome. This is true for sexual bacteria colonies that exchange plasmids. It is also true for animal species, e.g., variation in immune system DNA that protects the specie from pathogens. Variation is the fuel that selection burns for adaptation.
Eliezer: "Fly, you've just postulated four copies of the same gene, so that one death will remove four mutations. But these four copies will suffer mutations four times as often. Unless I'm missing something, this doesn't increase the bound on how much non-redundant information can be supported by one death. :)"
Yeah, you are right. You only gain if the redundancy means that the fitness hit is sufficiently minor that more than four errors could be removed with a single death.
The "one death, one mutation" rule applies if the mutation immediately affects the first generations. However, having backup copies means that mutations are seldom all that damaging. Humans have two copies of the genome (except for us poor males who suffer from X-linked genetic diseases). A loss-of-function mutation in a gene may have minor fitness impact. If a mutation causes failure to implant or an early miscarriage, then it should have little affect on the number of offspring a woman produces. If the mutation has minor fitness impact then the more efficient error correcting that occurs through crossover, chromosome competition, and mate competition could come into play.
Redundancy might increase the amount non-redundant information supported by one death, but not in the manner I presented in that example.
PS
In some cases assortative mating could also act to segregate beneficial and harmful alleles and accelerate filtering.
I like that evolution inherently prioritizes error removal. The worst mutations are removed quickly at a high "death" cost. Less harmful mutations are removed more slowly and at a lower "death" cost (since multiple "errors" are removed with each death).
I've been enjoying your evolution posts and wanted to toss in my own thoughts and see what I can learn.
"Our first lemma is a rule sometimes paraphrased as "one mutation, one death"."
Imagine that having a working copy of gene "E" is essential. Now suppose a mutation creates a broken gene "Ex". Animals that are heterozygous with "E" and "Ex" are fine and pass on their genes. Only homozygous "Ex" "Ex" result in a "death" that removes 2 mutations.
Now imagine that a duplication event gives four copies of "E". In this example an animal would only need one working gene out of the four possible copies. When the rare "Ex" "Ex" "Ex" "Ex" combination arises then the resulting "death" removes four mutations.
In fruit fly knock-out experiments, breaking one development gene often had no visible affect. Backup genes worked well enough. The backup gene could have multiple roles: First, it has a special function that improves the animal fitness. Second, it works as a backup when the primary gene is disabled. The resulting system is robust since the animal can thrive with many broken copies and evolution is efficient since a single "death" can remove four harmful mutations.
I've focussed on protein-coding genes, but this concept also applies to short DNA segments that code for elements such as miRNA's. Imagine that the DNA segment is duplicated. Being short, it is rarely deactivated by a mutation. Over time a genome may acquire many working copies that code for that miRNA. Rarely an animal would inherit no working copies and so a "death" would remove multiple chromosomes that "lacked" that DNA segment. On the other hand, too many copies might also be fatal. Chromosomes with too few or too many active copies would suffer a fitness penalty.
On a different note, imagine two stags. The first stag has lucked-out and inherited many alleles that improve its fitness. The second stag wasn't so lucky and inherited many bad alleles. The first stag successfully mates and the second doesn't. One "death" removed many inferior alleles.
Animals may have evolved sexual attraction based on traits that depend on the proper combined functioning of many genes. An unattractive mate might have many slightly harmful mutations. Thus one "death" based on sexual selection might remove many harmful mutations.
Evolution might be a little better than the "one mutation, one death" lemma implies. (I agree that evolution is an inefficient process.)
"This 1 bit per generation has to be divided up among all the genetic variants being selected on, for the whole population. It's not 1 bit per organism per generation, it's 1 bit per gene pool per generation."
Suppose new allele "A" has fitness advantage 1.03 compared to the wild allele "a" and that another allele "B" on the same type chromosome has fitness advantage 1.02. Eventually the "A" and "B" alleles will be sufficiently common that a crossover creating a new chromosome "AB" with "A" and "B" alleles is likely (This crossover probability depends on the population sizes of "Ab" and "aB" chromosomes and the distance between the alleles). The new chromosome "AB" should have a fitness of 1.05 compared to the chromosome "ab". Both "A" and "B" should then see an accelerated spread until the "ab" chromosomes are largely displaced. The rate would then diminish as "AB" displaced "Ab" and "aB" chromosomes. Thus multiple beneficial mutations of the same type chromosome should spread faster than the "single mutation" formula would indicate.
Due to crossover, good "bits" would tend to accumulate on good chromosomes thereby increasing the fitness of the entire chromosome as described above. The highly fit good chromosome thus displaces chromosome with many bad "bits". The good "bits" are no longer inherited independently and each "death" can now select multiple information "bits".
We seem to view evolution from a similar perspective.
Information requires selection in order to be preserved. The DNA information in an animal genome could be ranked in "fitness" value and the resulting graph would likely follow a power law. I.e., some DNA information is extremely important and likely to be preserved while most of the DNA is relatively free to drift. In a species such as fruit flies with many offspring selection can drive the species high up a local fitness peak. Much of the animal genome will be optimized. In a species such as humans with few offspring there is much less selection pressure and the specie gene pool wanders further from local peaks. More of the human genome drifts. (E.g., human regulatory elements are less conserved than rodent regulatory elements.)